CN112181659A - Cloud simulation memory resource prediction model construction method and memory resource prediction method - Google Patents
Cloud simulation memory resource prediction model construction method and memory resource prediction method Download PDFInfo
- Publication number
- CN112181659A CN112181659A CN202011071850.2A CN202011071850A CN112181659A CN 112181659 A CN112181659 A CN 112181659A CN 202011071850 A CN202011071850 A CN 202011071850A CN 112181659 A CN112181659 A CN 112181659A
- Authority
- CN
- China
- Prior art keywords
- model
- trained
- training
- cloud simulation
- memory resource
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5011—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resources being hardware resources other than CPUs, Servers and Terminals
- G06F9/5016—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resources being hardware resources other than CPUs, Servers and Terminals the resource being the memory
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to a cloud simulation memory resource prediction model construction method, a cloud simulation memory resource prediction model construction device, computer equipment and a storage medium, a random forest has fewer hyper-parameters needing to be adjusted, the prediction effect is good and very stable, a BP neural network can well represent a nonlinear relation and is noise-resistant, and Gaussian process regression can quantify uncertainty in prediction through probability prediction, so that the cloud simulation memory resource prediction model obtained by accumulating the three models can support accurate prediction of cloud simulation memory resources. In addition, the application also provides a method and a device for predicting the cloud simulation memory resources, computer equipment and a storage medium, which can realize accurate prediction of the cloud simulation memory resources.
Description
Technical Field
The application relates to the technical field of computers, in particular to a cloud simulation memory resource prediction model construction method and device, computer equipment, a storage medium, a memory resource prediction method and device, computer equipment and a storage medium.
Background
The large-scale complex system comprises a large number of components, and complex interaction exists among the components, such as combat countermeasures, economic models, disease transmission models and the like. The complex system simulation provides an effective method for researching the complex system. With the larger and larger simulation application scale of the complex system, the entity interaction becomes more and more complex, and higher requirements are put forward on the simulation performance. On the other hand, uncertainty of interaction between entities in the simulation application leads to varying resource requirements, and the traditional high-performance computing environment is difficult to support efficient operation of the complex system simulation application. The development of cloud computing provides a new solution for the deployment and operation of the simulation application of the complex system. The expandability and flexibility of the cloud computing provide dynamic expandable resources for the simulation application of the complex system, and the elastic resource requirements of the simulation application are met.
In a cloud environment, simulation entities included in a complex system simulation application are distributed to different groups, and frequent communication operations are generated due to interaction among the entities, which results in a large amount of memory allocation and recovery operations. Memory is one of the major factors limiting the performance of complex system simulation applications, and up to 60% of the time a processor spends waiting for a memory operation to complete. Particularly for simulation applications using optimistic time synchronization, if the allocated memory resources are too few, it is difficult to support efficient operation of the applications, and if the allocated memory resources are too many, on the one hand, resources are wasted, and on the other hand, the computational performance of the applications may be degraded, especially when the workload is unbalanced, because the limited use of the memory resources can avoid the applications from being excessively optimistically executed, thereby avoiding excessive rollback operations and performance degradation. The accurate allocation of the memory resources required by the simulation application has important significance for improving the performance of the simulation application and reducing the resource consumption.
It can be seen that the memory resource management and allocation of the complex system simulation application in the cloud environment is a challenging and open problem, and the memory resource needs to be accurately predicted for effective management and allocation of the memory resource. Therefore, a cloud simulation memory resource prediction scheme with accurate prediction is urgently needed at present.
Disclosure of Invention
Based on this, it is necessary to provide a method and an apparatus for constructing a cloud emulated memory resource prediction model, a computer device and a storage medium, which support accurate prediction of cloud emulated memory resources, in view of the above technical problems; and provides a cloud simulation memory resource prediction method, device, computer equipment and storage medium capable of accurately predicting.
A cloud simulation memory resource prediction model construction method comprises the following steps:
acquiring a sample set of memory data of cloud simulation application, wherein the sample set comprises a training set and a verification set;
acquiring a preset base model, training the preset base model by a training set to obtain a trained base model, and constructing the preset base model based on a BP neural network and a random forest;
inputting the input data in the verification set into the trained basic model to obtain a prediction result of the verification set;
generating a new training sample according to the prediction result of the verification set, and training a preset Gaussian process regression model according to the training sample to obtain a trained meta-model;
and constructing a cloud simulation memory resource prediction model according to the trained base model and the trained meta model.
In one embodiment, obtaining the pre-set base model, training the pre-set base model by the training set comprises:
acquiring a plurality of preset base models, and randomly dividing a training set into a plurality of training subsets;
training a single preset base model by adopting a single training subset to obtain a plurality of initial trained base models;
obtaining a prediction result root mean square error of each initially trained base model;
pruning the plurality of initial trained base models based on the root mean square error of the prediction result of each initial trained base model to obtain a trained base model set, wherein the trained base model set comprises a plurality of trained base models.
In one embodiment, pruning the plurality of initial trained base models based on the root mean square error of the prediction result of each initial trained base model to obtain the set of trained base models comprises:
acquiring the number N of initial trained base models;
sequencing each initially trained base model according to the root mean square error of the corresponding prediction result to obtain a sequencing queue;
sequentially selecting initial trained base models with different numbers i according to the sorting queue, and calculating the root mean square error until the number i reaches the number N, wherein N is more than or equal to 2;
and selecting the base model set corresponding to the minimum root mean square error to obtain the trained base model set.
In one embodiment, randomly dividing the training set into a plurality of training subsets comprises:
and sampling the training set by adopting a bootstrap method to obtain a plurality of training subsets.
In one embodiment, obtaining a sample set of cloud simulation application memory data comprises:
acquiring a memory data set of simulation application deployed in a cloud environment according to a preset time window;
searching abnormal values in the memory data set by adopting a boxplot mode;
and positioning the abnormal values based on the time sequence, and replacing the abnormal values with the upper sample and the lower sample corresponding to the abnormal values based on the time sequence to obtain a sample set of the memory data of the cloud simulation application.
The application also provides a cloud simulation memory resource prediction model construction device, which comprises:
the cloud simulation system comprises a sample acquisition module, a verification module and a data processing module, wherein the sample acquisition module is used for acquiring a sample set of memory data of cloud simulation application, and the sample set comprises a training set and a verification set;
the base model training module is used for acquiring a preset base model, training the preset base model by a training set to obtain a trained base model, and constructing the preset base model based on a BP neural network and a random forest;
the verification module is used for inputting the input data in the verification set into the trained base model to obtain a prediction result of the verification set;
the meta-model training module is used for generating a new training sample according to the verification set prediction result and training a preset Gaussian process regression model according to the training sample to obtain a trained meta-model;
and the model construction module is used for constructing a cloud simulation memory resource prediction model according to the trained base model and the trained meta model.
The present application further provides a computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method as described above when executing the computer program.
The present application also provides a computer-readable storage medium having stored thereon a computer program which, when being executed by a processor, carries out the steps of the method as described above.
The cloud simulation memory resource prediction model construction method, the cloud simulation memory resource prediction model construction device, the computer equipment and the storage medium acquire a sample set of memory data applied to cloud simulation, and train a base model by adopting a training set with models constructed by a BP neural network and a random forest as the base model; and (3) taking the Gaussian regression model as a meta model, generating a training sample of the meta model by using data output by the base model under the verification set to obtain the trained meta model, and combining the base model and the remote model to obtain a final cloud simulation memory resource prediction model. In the whole process, the random forest has fewer hyper-parameters needing to be adjusted, the prediction effect is good and stable, the BP neural network can well represent the nonlinear relation and is noise-resistant, and the Gaussian process regression can quantify uncertainty in prediction through probability prediction, so that the cloud simulation memory resource prediction model obtained by accumulating the three models can support accurate prediction of cloud simulation memory resources.
In addition, the present application also provides a cloud simulation memory resource prediction method, including:
collecting cloud simulation application state information;
inputting the cloud simulation application state information into a trained cloud simulation memory resource prediction model, wherein the trained cloud simulation memory resource prediction model is constructed by the method;
and acquiring output data of the trained cloud simulation memory resource prediction model to obtain a cloud simulation application memory resource prediction result.
In one embodiment, the cloud simulation memory resource prediction method further includes:
monitoring memory resources allocated to the cloud simulation application;
and feeding back and updating the trained cloud simulation memory resource prediction model according to the predicted cloud simulation application memory resource prediction result at the last moment and the memory resource of the cloud simulation application at the current moment.
In addition, the present application further provides a cloud emulation memory resource prediction apparatus, including:
the data acquisition module is used for cloud simulation application state information;
the data input module is used for inputting the cloud simulation application state information into the trained cloud simulation memory resource prediction model, and the trained cloud simulation memory resource prediction model is constructed by the method;
and the prediction module is used for acquiring the output data of the trained cloud simulation memory resource prediction model to obtain a cloud simulation application memory resource prediction result.
A computer device comprising a memory storing a computer program and a processor implementing the steps of the method as described above when the computer program is executed by the processor.
A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method as described above.
According to the cloud simulation memory resource prediction method, the cloud simulation memory resource prediction device, the computer equipment and the storage medium, the cloud simulation application memory resource prediction is carried out based on the trained cloud simulation memory resource prediction model, the trained cloud simulation memory resource prediction model is obtained based on three networks of BP neural network, random forest and Gaussian linear regression through stacking training, the number of the hyper-parameters needing to be adjusted in the random forest is small, the prediction effect is good and stable, the BP neural network can well represent the nonlinear relation and is noise-resistant, and the Gaussian process regression can quantify the uncertainty in the prediction through probability prediction, so that the accurate prediction of the cloud simulation memory resources can be realized.
Drawings
Fig. 1 is an application environment diagram of a cloud simulation memory resource prediction model construction method in an embodiment;
FIG. 2 is a schematic flow chart illustrating a method for building a cloud-emulated memory resource prediction model in an embodiment;
FIG. 3 is a schematic diagram of a BP neural network structure;
FIG. 4 is a schematic diagram of a random forest flow;
FIG. 5 is a schematic flow chart illustrating a method for building a cloud emulated memory resource prediction model in another embodiment;
FIG. 6 is a flow chart of a pruning algorithm based on root mean square error;
FIG. 7 is a schematic flow chart of a method for building a cloud simulation memory resource prediction model in an application example;
fig. 8 is a schematic structural diagram of a cloud simulation memory resource prediction model construction device in one embodiment;
fig. 9 is a schematic flowchart illustrating a method for predicting cloud-emulated memory resources in an embodiment;
fig. 10 is a schematic structural diagram of a cloud emulated memory resource prediction device in an embodiment;
FIG. 11 is a diagram illustrating an architecture of a cloud emulated memory resource prediction device in an application example;
FIG. 12 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The cloud simulation memory resource prediction model construction method provided by the application can be applied to the application environment shown in fig. 1. The server 102 is connected with the cloud system 104 through a network, the server 102 collects cloud simulation application memory data in the cloud system 104 in historical operation, and the server 102 takes the data as a sample set which comprises a training set and a verification set; the server 102 reads a pre-stored base model constructed based on a BP neural network and a random forest, and trains the pre-stored base model by adopting a sample set; inputting the input data in the verification set into the trained basic model to obtain a prediction result of the verification set; generating a new training sample according to the prediction result of the verification set, and training a preset Gaussian process regression model according to the training sample to obtain a trained meta-model; and constructing a cloud simulation memory resource prediction model according to the trained base model and the trained meta model. The server 102 may be a monitoring server device for monitoring the operation of the cloud system, and may be implemented by a stand-alone server or a server cluster composed of a plurality of servers. It can be understood that the cloud simulation memory resource prediction model building method can also be directly executed by a cloud system.
In addition, the present application also provides a cloud simulation memory resource prediction method, which can also be applied to the application environment shown in fig. 1, the server 102 loads a cloud simulation memory resource prediction model based on the cloud simulation memory resource prediction model construction method, and the server 102 collects cloud simulation application state information by monitoring a cloud system; and inputting the cloud simulation application state information into the trained cloud simulation memory resource prediction model, and acquiring the output data of the trained cloud simulation memory resource prediction model to obtain a cloud simulation application memory resource prediction result. It is to be understood that the cloud emulation memory resource prediction method described above may also be directly executed by the cloud system.
In an embodiment, as shown in fig. 2, a method for building a cloud simulation memory resource prediction model is provided, which is described by taking the method as an example applied to the server 102 in fig. 1, and includes the following steps:
s210: and acquiring a sample set of memory data of the cloud simulation application, wherein the sample set comprises a training set and a verification set.
The sample set of the cloud simulation application memory data may be a sample set obtained by collecting cloud simulation application state data and required memory resources in a history, and may also be standard sample data given by a third party. Specifically, if a historical data acquisition mode is adopted, the simulation application can be deployed and operated in a cloud environment (cloud system), cloud simulation application state data and required memory resources are collected by taking 10s (which can be adjusted according to actual needs) as a window, and a sample set can be obtained after the collection for a period of time. The sample set includes a training set and a validation set, wherein the training set is used for training the model, and the validation set is used for validating and testing the model, generally, the data amount of the training set is greater than that of the validation set, and specifically, 80% of the sample set can be selected as the training set, and 20% of the sample set can be selected as the validation set.
S220: and acquiring a preset base model, training the preset base model by a training set to obtain the trained base model, and constructing the preset base model based on the BP neural network and the random forest.
The preset base model is constructed based on a BP neural network and a random forest, and can be understood that the model has two parts of the BP neural network and the random forest which are parallel, and the two parts are trained by adopting a training set respectively during training.
Specifically, the BP neural network generally refers to a multi-layer feedforward neural network to which a BP algorithm is applied, and is excellent in solving a regression problem. The structure of the BP neural network is shown in fig. 3, and a typical BP neural network is composed of three layers, an input layer, a hidden layer, and an output layer. For each training sample, the BP neural network calculates the error of the output layer, then reversely propagates the error to the hidden layer, finally adjusts the weight and the threshold value according to the error, and can reduce the training error through continuous iteration. And integrating the decision tree by the random forest by using bagging, and further adding random feature selection in the training of the decision tree. In other words, each independent decision tree is trained on randomly selected samples and randomly selected features. The random forest increases the difference of the base model through sample disturbance and characteristic disturbance, so that the generalization capability of the random forest is further improved. And thus are widely used in classification and regression problems. For the regression problem, the mean of the decision tree set is used as the output of the random forest, and the random forest regression flow is shown in fig. 4.
S230: and inputting the input data in the verification set into the trained basic model to obtain a prediction result of the verification set.
Having obtained the trained base model at S220, the validation set input data is input to the trained base model for the desired resource prediction. As described above, the verification set belongs to a part of the sample set, and in brief, the verification set is data including two parts, namely, cloud simulation application state data (X data for short) and required memory resources (Y data for short), wherein the cloud simulation application state data belongs to input data thereof, and the required memory resources belong to output data thereof, and the cloud simulation application state data (X data) in the verification set is input to the trained base model to obtain a verification set prediction result, so as to obtain a predicted memory resource (Y1 data for short).
S240: and generating a new training sample according to the prediction result of the verification set, and training a preset Gaussian process regression model according to the training sample to obtain a trained meta-model.
And (3) generating a new training sample by adopting the verification set prediction result obtained in the step (S230) to train the preset Gaussian process regression model, namely replacing the input data (X data) in the verification set with the predicted memory resource (Y1 data) to obtain a new training sample which comprises the predicted memory resource data (Y1 data) output by the trained base model and the memory resource data (Y data) in the verification set, wherein the new training sample can be simply understood as a new training sample obtained by (Y1, Y). And training a preset Gaussian process regression model by using the new training sample to obtain a trained meta-model. The process relates to a stacking integration mode, and a base model constructed based on a BP neural network and a random forest is connected with a meta model based on a Gaussian process regression model by the stacking integration mode. Specifically, the sample set would be divided into 80% training set and 20% validation set. The first layer basis model performs 5-fold cross validation, the first layer basis model outputs the prediction result of the validation set, and 5 times of cross validation can obtain the memory resource prediction result corresponding to 5 × 20% of the validation set, where the result is that the output of the first layer basis model is used as the input x of the gaussian process regression, and then y of the gaussian process regression is the y of the data validation set, which is equivalent to replacing x in the validation set with the output of the first layer basis model, and y is unchanged.
Specifically, the gaussian process regression is a probabilistic predictive model under the bayesian framework, and the properties of the model are represented by a mean function m (x) and a covariance function cov (x, x'). The details are as follows:
f(x)～GP(m(x)，cov(x，x′))
m(x)＝E[f(x)]
cov(x，x′)＝E[(f(x)-m(x))*(f(x′)-m(x′))]
in an actual regression problem, the influence of noise needs to be considered, so the predicted value considering noise can be expressed as:
y＝f(x)+
～N(0，σ2)
consider a test sample xtThe predicted value is f (x)t) Then f (x)t) The joint distribution with the training set observations y can be expressed as:
k＝cov(x，x)+σ2I
f (x) is obtained from the above formulat) The posterior distribution of (a) can be expressed as:
f(xt)|xy，xt～N(f(xt)′，cov(f(xt)))
f(xt)′＝cov(xt，x)k-1y
cov(f(xt))＝cov(xt，xt)-cov(xt，x)k-1cov(x，xt)
f(xt) ' is a Gaussian process regression predictor, cov (f (x)t) Is the predicted value variance of the gaussian process regression. Therefore, gaussian process regression is often used as a probabilistic predictive model.
S250: and constructing a cloud simulation memory resource prediction model according to the trained base model and the trained meta model.
And integrating the trained base model and the trained meta model to construct a cloud simulation memory resource prediction model. Three models, namely a BP neural network, a random forest and Gaussian process regression, are integrated in the cloud simulation memory resource prediction model, and the cloud simulation memory resources are accurately predicted through the model in subsequent application. Optionally, after the cloud simulation memory resource prediction model is obtained, the cloud simulation memory resource prediction model may be tested and verified by using data to further ensure accuracy of subsequent use, specifically, after sample data is obtained, the sample data may be divided into a sample set and an additional test set, and 80% of the sample data may be used as the sample set and 20% of the sample data may be used as the additional test set.
The cloud simulation memory resource prediction model construction method comprises the steps of obtaining a sample set of memory data of cloud simulation application, taking a model constructed by a BP neural network and a random forest as a base model, and training the base model by adopting a training set; and (3) taking the Gaussian regression model as a meta model, generating a training sample of the meta model by using data output by the base model under the verification set to obtain the trained meta model, and combining the base model and the remote model to obtain a final cloud simulation memory resource prediction model. In the whole process, the random forest has fewer hyper-parameters needing to be adjusted, the prediction effect is good and stable, the BP neural network can well represent the nonlinear relation and is noise-resistant, and the Gaussian process regression can quantify uncertainty in prediction through probability prediction, so that the cloud simulation memory resource prediction model obtained by accumulating the three models can support accurate prediction of cloud simulation memory resources.
As shown in fig. 5, in one embodiment, S220 includes:
s222: acquiring a plurality of preset base models, and randomly dividing a training set into a plurality of training subsets;
s224: training a single preset base model by adopting a single training subset to obtain a plurality of initial trained base models;
s226: obtaining a prediction result root mean square error of each initially trained base model;
s228: pruning the plurality of initial trained base models based on the root mean square error of the prediction result of each initial trained base model to obtain a trained base model set, wherein the trained base model set comprises a plurality of trained base models.
In order to improve the performance of the integrated model, an optimal subset of the base models needs to be selected to eliminate the influence of the base models with poor effects, and in this embodiment, a pruning algorithm based on RMSE (root mean square error) is proposed to screen out a base model set obtained by training the optimal subset, where the trained base model set includes a plurality of trained base models. Specifically, firstly, a certain number of preset base models are selected, the preset base models are all the same initial models, a training set is divided into a plurality of training subsets in a random mode, data contained in the training subsets are different, the training subsets can be sampled by a bootstrap method to increase disturbance of samples to obtain a plurality of training subsets, the number of the training subsets can be specifically the same as that of the preset base models, a single training subset is adopted to train a single preset base model to obtain a plurality of initial trained base models, RMSE of the initial trained base models is obtained, specifically, the same group of test sample data can be input into the plurality of initial trained base models, RMSE of the initial trained base models is obtained through calculation, the group of data can be data extracted from training data or additionally imported sample data, the specific calculation formula is as follows:
in the formula, yiAnd ypredictioniRespectively representing the true value and the predicted value of the sample i; n is the number of samples in the set. After the root mean square error of the prediction result of each initial trained base model is obtained, pruning is carried out on a plurality of initial trained base models, and finally a trained base model set is obtained.
In one embodiment, the pruning to obtain the trained set of base models includes: acquiring the number N of initial trained base models; sequencing each initially trained base model according to the root mean square error of the corresponding prediction result to obtain a sequencing queue; sequentially selecting initial trained base models with different numbers i according to the sorting queue, and calculating the root mean square error until the number i reaches the number N, wherein N is more than or equal to 2; and selecting the base model set corresponding to the minimum root mean square error to obtain the trained base model set.
Briefly, the pruning process described above requires that the root mean square error of the set of base models consisting of a plurality of trained base models be selected to be minimal. Specifically, as shown in fig. 6, the number N of initial trained base models is obtained after previous processing is obtained, each initial trained base model is sorted in an ascending or descending manner according to the root mean square error of the corresponding prediction result (preferably, an ascending manner is selected), a sorting queue is obtained, different numbers i of initial trained base models are sequentially selected from the sorting queue to form a base model set, the root mean square errors in the base model sets are respectively calculated, it can be understood that i is increased one by one from 2, that is, i can sequentially take values 2, 3, 4, … …, and N to obtain a base model set including different numbers of initial trained base models, the root mean square error of each base model set is calculated, and the corresponding base model set with the smallest root mean square error is selected to obtain a trained base model set.
In one embodiment, obtaining a sample set of cloud simulation application memory data comprises: acquiring a memory data set of simulation application deployed in a cloud environment according to a preset time window; searching abnormal values in the memory data set by adopting a boxplot mode; and positioning the abnormal values based on the time sequence, and replacing the abnormal values with the upper sample and the lower sample corresponding to the abnormal values based on the time sequence to obtain a sample set of the memory data of the cloud simulation application.
In this embodiment, an abnormal value in the memory data set is searched in a boxplot manner, and a previous sample and a next sample corresponding to the abnormal value are used to replace the abnormal value, so that the abnormal value is cleaned and corrected, and the obtained sample set is more accurate.
Specifically, referring to fig. 7 for one application example of the cloud simulation memory resource prediction model building method, as shown in fig. 7, in order to accurately predict memory resources required by a simulation application, the simulation application is deployed and operated in a cloud environment, dynamic information of the simulation application is collected by taking 10s as a window, and static information and the dynamic information are combined to generate a simulation application memory data set. The collected data is then cleaned up and we use a boxplot to find outliers in the data set, which are replaced by the mean of the previous and next sample. Meanwhile, in order to eliminate the influence of the dimension on the result, all data are subjected to standardization processing. The simulation uses 80% of samples of the memory data set as a data training set, 20% of samples of the memory data set as a data testing set, the data training set is used as a data set of the first-layer base model, 80% of the data training set is used as a training set, and 20% of the data training set is used as a verification set. The data test set is used as a test set of the memory resource prediction model. By integrating a plurality of base models, the memory prediction accuracy of the simulation application of the complex system can be improved, and the problem of poor generalization capability of a single model can be solved. Therefore, the BP neural network, the random forest and the Gaussian process regression are integrated in a stacking mode. The performance of the integrated model mainly includes two factors, the difference of the base model and the performance of the base model. To improve the difference between the base models, we use bootstrap sampling to increase the disturbance of the sample, and to improve the performance of the integrated model, we need to select the optimal subset of the base models to eliminate the influence of the less effective base models. Therefore, we propose a pruning algorithm based on RMSE. The algorithm can select an optimal subset of the basis models based on the RMSE index. And the output of the selected optimal base model is used as the input of Gaussian process regression, then the Gaussian process regression is used as the memory resource required by the second-layer meta-model prediction simulation application, the neural network, the random forest and the Gaussian process regression are combined to construct a cloud simulation memory resource prediction model, and the cloud simulation memory resource prediction model can accurately predict the memory resource required by the simulation application.
As shown in fig. 8, the present application further provides a cloud simulation memory resource prediction model building apparatus, which includes:
the sample acquisition module 810 is configured to acquire a sample set of memory data of the cloud simulation application, where the sample set includes a training set and a verification set;
a base model training module 820, configured to obtain a preset base model, train the preset base model by a training set to obtain a trained base model, where the preset base model is constructed based on a BP neural network and a random forest;
the verification module 830 is configured to input the input data in the verification set to the trained base model to obtain a prediction result of the verification set;
the meta-model training module 840 is used for generating a new training sample according to the verification set prediction result and training a preset Gaussian process regression model according to the training sample to obtain a trained meta-model;
and the model building module 850 is used for building a cloud simulation memory resource prediction model according to the trained base model and the trained meta model.
The cloud simulation memory resource prediction model construction device acquires a sample set of memory data applied to cloud simulation, takes a model constructed by a BP neural network and a random forest as a base model, and trains the base model by adopting a training set; and (3) taking the Gaussian regression model as a meta model, generating a training sample of the meta model by using data output by the base model under the verification set to obtain the trained meta model, and combining the base model and the remote model to obtain a final cloud simulation memory resource prediction model. In the whole process, the random forest has fewer hyper-parameters needing to be adjusted, the prediction effect is good and stable, the BP neural network can well represent the nonlinear relation and is noise-resistant, and the Gaussian process regression can quantify uncertainty in prediction through probability prediction, so that the cloud simulation memory resource prediction model obtained by accumulating the three models can support accurate prediction of cloud simulation memory resources.
In one embodiment, the base model training module 820 is further configured to obtain a plurality of preset base models, and randomly divide the training set into a plurality of training subsets; training a single preset base model by adopting a single training subset to obtain a plurality of initial trained base models; obtaining a prediction result root mean square error of each initially trained base model; pruning the plurality of initial trained base models based on the root mean square error of the prediction result of each initial trained base model to obtain a trained base model set, wherein the trained base model set comprises a plurality of trained base models.
In one embodiment, the base model training module 820 is further configured to obtain a number N of initial trained base models; sequencing each initially trained base model according to the root mean square error of the corresponding prediction result to obtain a sequencing queue; sequentially selecting initial trained base models with different numbers i according to the sorting queue, and calculating the root mean square error until the number i reaches the number N, wherein N is more than or equal to 2; and selecting the base model set corresponding to the minimum root mean square error to obtain the trained base model set.
In one embodiment, the base model training module 820 is further configured to sample the training set using a bootstrap method to obtain a plurality of training subsets.
In one embodiment, the sample obtaining module 810 is further configured to collect, according to a preset time window, a memory data set of the simulation application deployed in the cloud environment; searching abnormal values in the memory data set by adopting a boxplot mode; and positioning the abnormal values based on the time sequence, and replacing the abnormal values with the upper sample and the lower sample corresponding to the abnormal values based on the time sequence to obtain a sample set of the memory data of the cloud simulation application.
In addition, as shown in fig. 9, the present application further provides a cloud simulation memory resource prediction method, including:
s920: collecting cloud simulation application state information;
s940: inputting the cloud simulation application state information into a trained cloud simulation memory resource prediction model, wherein the trained cloud simulation memory resource prediction model is constructed by the method;
s960: and acquiring output data of the trained cloud simulation memory resource prediction model to obtain a cloud simulation application memory resource prediction result.
According to the cloud simulation memory resource prediction method, the cloud simulation application memory resource prediction is carried out based on the trained cloud simulation memory resource prediction model, the trained cloud simulation memory resource prediction model is obtained based on three networks of BP neural network, random forest and Gaussian linear regression through stacking training, the number of hyper-parameters needing to be adjusted in the random forest is small, the prediction effect is good and stable, the BP neural network can well represent the nonlinear relation and is noise-resistant, and the Gaussian process regression can quantify the uncertainty in the prediction through probability prediction, so that the accurate prediction of the cloud simulation memory resources can be realized.
In one embodiment, the cloud simulation memory resource prediction method further includes:
monitoring memory resources allocated to the cloud simulation application; and feeding back and updating the trained cloud simulation memory resource prediction model according to the predicted cloud simulation application memory resource prediction result at the last moment and the memory resource of the cloud simulation application at the current moment.
In this embodiment, an iterative update mechanism is further introduced for the cloud-emulated memory resource prediction, that is, the predicted value at the previous time is compared with the memory resource of the cloud-emulated application at the current time for feedback, and a feedback update mechanism is adopted to update the trained cloud-emulated memory resource prediction model, so that the cloud-emulated memory resource prediction model prediction is closer to the true value, and the prediction of the whole cloud-emulated memory resource prediction scheme is more and more accurate.
In addition, as shown in fig. 10, the present application further provides a cloud emulation memory resource prediction apparatus, including:
the data acquisition module 970 is used for acquiring the state information of the cloud simulation application;
a data input module 980, configured to input cloud simulation application state information to the trained cloud simulation memory resource prediction model, where the trained cloud simulation memory resource prediction model is constructed by the above method;
the prediction module 990 is configured to obtain data output by the trained cloud simulation memory resource prediction model, and obtain a cloud simulation application memory resource prediction result.
According to the cloud simulation memory resource prediction device, the cloud simulation application memory resource prediction is carried out based on the trained cloud simulation memory resource prediction model, the trained cloud simulation memory resource prediction model is obtained based on three networks of BP neural network, random forest and Gaussian linear regression through stacking training, the number of hyper-parameters needing to be adjusted in the random forest is small, the prediction effect is good and stable, the BP neural network can well represent the nonlinear relation and is noise-resistant, and the Gaussian process regression can quantify the uncertainty in the prediction through probability prediction, so that the accurate prediction of the cloud simulation memory resources can be realized.
As shown in fig. 11, the cloud simulation memory resource prediction apparatus may specifically include a cloud resource module, a data set module, and a memory resource prediction model module.
The cloud resource module is mainly used for monitoring the simulation application state and distributing memory resources required by the simulation application. The monitor monitors the running state of the simulation application in real time and returns monitoring information to the data set module. The distributor distributes the simulation application to the corresponding computing nodes according to the memory resources predicted by the memory prediction model and the available cloud resources.
The data set module collects and preprocesses the static and dynamic information of the simulation application. The static information comprises the number of simulation entities and simulation end time information. The dynamic information comprises simulation application running information and cloud resource information. The simulation application running information comprises CPU utilization rate, memory utilization rate, simulation event rollback number, simulation running time, network time delay and network receiving/sending messages. The data set module can carry out standardization processing on the collected data so as to eliminate dimensional inconsistency and provide training and testing for the memory resource prediction model.
The memory resource prediction model trains the integrated model by using historical data, namely a feedback loop is formed, and the problem that the memory resource prediction data quantity is insufficient is solved. We choose a stacking integration strategy to accomplish the prediction of memory resources, which includes two layers of models. The performance of the integrated model is related to the diversity of each base model, in order to increase the diversity among the base models, a random forest and a BP neural network are selected as a first layer base model, and Gaussian process regression is used as a second layer base model to generate final probability prediction. The provided integrated model can accurately predict the memory resources required by the simulation application. And finally, outputting the predicted result to the cloud resource module.
It should be understood that, although the steps in the above-described flowcharts are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in the flowcharts shown above may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of performing the steps or stages is not necessarily sequential, but may be performed alternately or alternately with other steps or at least a portion of the steps or stages in other steps.
For specific limitations of the cloud simulation memory resource prediction model construction device, reference may be made to the above limitations of the cloud simulation memory resource prediction model construction method, which is not described herein again. All modules in the cloud simulation memory resource prediction model building device can be wholly or partially realized through software, hardware and a combination of the software and the hardware. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 12. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer equipment is used for storing historical sample data or preset base models and other data. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a cloud emulated memory resource prediction model construction method or a cloud emulated memory resource prediction method.
Those skilled in the art will appreciate that the architecture shown in fig. 12 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
acquiring a sample set of memory data of cloud simulation application, wherein the sample set comprises a training set and a verification set;
acquiring a preset base model, training the preset base model by a training set to obtain a trained base model, and constructing the preset base model based on a BP neural network and a random forest;
inputting the input data in the verification set into the trained basic model to obtain a prediction result of the verification set;
generating a new training sample according to the prediction result of the verification set, and training a preset Gaussian process regression model according to the training sample to obtain a trained meta-model;
and constructing a cloud simulation memory resource prediction model according to the trained base model and the trained meta model.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
acquiring a plurality of preset base models, and randomly dividing a training set into a plurality of training subsets; training a single preset base model by adopting a single training subset to obtain a plurality of initial trained base models; obtaining a prediction result root mean square error of each initially trained base model; pruning the plurality of initial trained base models based on the root mean square error of the prediction result of each initial trained base model to obtain a trained base model set, wherein the trained base model set comprises a plurality of trained base models.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
acquiring the number N of initial trained base models; sequencing each initially trained base model according to the root mean square error of the corresponding prediction result to obtain a sequencing queue; sequentially selecting initial trained base models with different numbers i according to the sorting queue, and calculating the root mean square error until the number i reaches the number N, wherein N is more than or equal to 2; and selecting the base model set corresponding to the minimum root mean square error to obtain the trained base model set.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
and sampling the training set by adopting a bootstrap method to obtain a plurality of training subsets.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
acquiring a memory data set of simulation application deployed in a cloud environment according to a preset time window; searching abnormal values in the memory data set by adopting a boxplot mode; and positioning the abnormal values based on the time sequence, and replacing the abnormal values with the upper sample and the lower sample corresponding to the abnormal values based on the time sequence to obtain a sample set of the memory data of the cloud simulation application.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
collecting cloud simulation application state information;
inputting the cloud simulation application state information into a trained cloud simulation memory resource prediction model, wherein the trained cloud simulation memory resource prediction model is constructed by the method;
and acquiring output data of the trained cloud simulation memory resource prediction model to obtain a cloud simulation application memory resource prediction result.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
monitoring memory resources allocated to the cloud simulation application; and feeding back and updating the trained cloud simulation memory resource prediction model according to the predicted cloud simulation application memory resource prediction result at the last moment and the memory resource of the cloud simulation application at the current moment.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring a sample set of memory data of cloud simulation application, wherein the sample set comprises a training set and a verification set;
acquiring a preset base model, training the preset base model by a training set to obtain a trained base model, and constructing the preset base model based on a BP neural network and a random forest;
inputting the input data in the verification set into the trained basic model to obtain a prediction result of the verification set;
generating a new training sample according to the prediction result of the verification set, and training a preset Gaussian process regression model according to the training sample to obtain a trained meta-model;
and constructing a cloud simulation memory resource prediction model according to the trained base model and the trained meta model.
In one embodiment, the computer program when executed by the processor further performs the steps of:
acquiring a plurality of preset base models, and randomly dividing a training set into a plurality of training subsets; training a single preset base model by adopting a single training subset to obtain a plurality of initial trained base models; obtaining a prediction result root mean square error of each initially trained base model; pruning the plurality of initial trained base models based on the root mean square error of the prediction result of each initial trained base model to obtain a trained base model set, wherein the trained base model set comprises a plurality of trained base models.
In one embodiment, the computer program when executed by the processor further performs the steps of:
acquiring the number N of initial trained base models; sequencing each initially trained base model according to the root mean square error of the corresponding prediction result to obtain a sequencing queue; sequentially selecting initial trained base models with different numbers i according to the sorting queue, and calculating the root mean square error until the number i reaches the number N, wherein N is more than or equal to 2; and selecting the base model set corresponding to the minimum root mean square error to obtain the trained base model set.
In one embodiment, the computer program when executed by the processor further performs the steps of:
and sampling the training set by adopting a bootstrap method to obtain a plurality of training subsets.
In one embodiment, the computer program when executed by the processor further performs the steps of:
acquiring a memory data set of simulation application deployed in a cloud environment according to a preset time window; searching abnormal values in the memory data set by adopting a boxplot mode; and positioning the abnormal values based on the time sequence, and replacing the abnormal values with the upper sample and the lower sample corresponding to the abnormal values based on the time sequence to obtain a sample set of the memory data of the cloud simulation application.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
collecting cloud simulation application state information;
inputting the cloud simulation application state information into a trained cloud simulation memory resource prediction model, wherein the trained cloud simulation memory resource prediction model is constructed by the method;
and acquiring output data of the trained cloud simulation memory resource prediction model to obtain a cloud simulation application memory resource prediction result.
In one embodiment, the computer program when executed by the processor further performs the steps of:
monitoring memory resources allocated to the cloud simulation application; and feeding back and updating the trained cloud simulation memory resource prediction model according to the predicted cloud simulation application memory resource prediction result at the last moment and the memory resource of the cloud simulation application at the current moment.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above examples only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A cloud simulation memory resource prediction model construction method is characterized by comprising the following steps:
acquiring a sample set of memory data of cloud simulation application, wherein the sample set comprises a training set and a verification set;
acquiring a preset base model, training the preset base model by the training set to obtain a trained base model, wherein the preset base model is constructed based on a BP neural network and a random forest;
inputting the input data in the verification set into the trained basic model to obtain a verification set prediction result;
generating a new training sample according to the prediction result of the verification set, and training a preset Gaussian process regression model according to the training sample to obtain a trained meta-model;
and constructing a cloud simulation memory resource prediction model according to the trained base model and the trained meta model.
2. The method of claim 1, wherein the obtaining a pre-set base model, the training the pre-set base model from the training set, and the obtaining the trained base model comprises:
obtaining a plurality of preset base models, and randomly dividing the training set into a plurality of training subsets;
training the single preset base model by adopting a single training subset to obtain a plurality of initial trained base models;
obtaining a prediction result root mean square error of each initially trained base model;
pruning the plurality of initial trained base models based on the root mean square error of the prediction result of each initial trained base model to obtain a trained base model set, wherein the trained base model set comprises a plurality of trained base models.
3. The method of claim 2, wherein pruning the plurality of initial trained base models based on the root mean square error of the prediction for each initial trained base model to obtain a set of trained base models comprises:
acquiring the number N of initial trained base models;
sequencing each initially trained base model according to the root mean square error of the corresponding prediction result to obtain a sequencing queue;
sequentially selecting initial trained base models with different numbers i according to the sorting queue, and calculating the root mean square error until the number i reaches the number N, wherein N is more than or equal to 2;
and selecting the base model set corresponding to the minimum root mean square error to obtain the trained base model set.
4. The method of claim 2, wherein the randomly partitioning the training set into a plurality of training subsets comprises:
and sampling the training set by adopting a bootstrap method to obtain a plurality of training subsets.
5. The method of claim 1, wherein obtaining a sample set of cloud simulation application memory data comprises:
acquiring a memory data set of simulation application deployed in a cloud environment according to a preset time window;
searching abnormal values in the memory data set in a boxplot mode;
and positioning the abnormal values based on the time sequence, and replacing the abnormal values with the previous sample and the next sample corresponding to the abnormal values based on the time sequence to obtain a sample set of the memory data of the cloud simulation application.
6. A cloud simulation memory resource prediction method is characterized by comprising the following steps:
collecting cloud simulation application state information;
inputting the cloud simulation application state information into a trained cloud simulation memory resource prediction model, wherein the trained cloud simulation memory resource prediction model is constructed by the method of any one of claims 1 to 5;
and acquiring the output data of the trained cloud simulation memory resource prediction model to obtain a cloud simulation application memory resource prediction result.
7. The method of claim 6, further comprising:
monitoring memory resources allocated to the cloud simulation application;
and feeding back and updating the trained cloud simulation memory resource prediction model according to the predicted cloud simulation application memory resource prediction result at the last moment and the memory resource of the cloud simulation application at the current moment.
8. A cloud simulation memory resource prediction model construction device is characterized by comprising the following components:
the cloud simulation system comprises a sample acquisition module, a verification module and a data processing module, wherein the sample acquisition module is used for acquiring a sample set of memory data of cloud simulation application, and the sample set comprises a training set and a verification set;
the base model training module is used for acquiring a preset base model, training the preset base model by the training set to obtain a trained base model, and constructing the preset base model based on a BP neural network and a random forest;
the verification module is used for inputting the input data in the verification set into the trained base model to obtain a verification set prediction result;
the meta-model training module is used for generating a new training sample according to the verification set prediction result and training a preset Gaussian process regression model according to the training sample to obtain a trained meta-model;
and the model construction module is used for constructing a cloud simulation memory resource prediction model according to the trained base model and the trained meta model.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 7.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011071850.2A CN112181659B (en) | 2020-10-09 | 2020-10-09 | Cloud simulation memory resource prediction model construction method and memory resource prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011071850.2A CN112181659B (en) | 2020-10-09 | 2020-10-09 | Cloud simulation memory resource prediction model construction method and memory resource prediction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112181659A true CN112181659A (en) | 2021-01-05 |
| CN112181659B CN112181659B (en) | 2023-07-18 |
Family
ID=73948901
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011071850.2A Active CN112181659B (en) | 2020-10-09 | 2020-10-09 | Cloud simulation memory resource prediction model construction method and memory resource prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112181659B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113268350A (en) * | 2021-06-07 | 2021-08-17 | 上海数禾信息科技有限公司 | Resource allocation method and device based on cloud service construction and computer equipment |
| CN113362218A (en) * | 2021-05-21 | 2021-09-07 | 北京百度网讯科技有限公司 | Data processing method and device, electronic equipment and storage medium |
| CN113688989A (en) * | 2021-08-31 | 2021-11-23 | 中国平安人寿保险股份有限公司 | Deep learning network acceleration method, device, equipment and storage medium |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10788900B1 (en) * | 2017-06-29 | 2020-09-29 | Snap Inc. | Pictorial symbol prediction |
| CN109240929A (en) * | 2018-09-18 | 2019-01-18 | 百度在线网络技术（北京）有限公司 | Software quality prediction method, apparatus, terminal and computer readable storage medium |
| CN110333991B (en) * | 2019-05-30 | 2022-11-25 | 武汉科技大学 | Method for predicting maximum resource utilization rate of cloud platform tasks |
-
2020
- 2020-10-09 CN CN202011071850.2A patent/CN112181659B/en active Active
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113362218A (en) * | 2021-05-21 | 2021-09-07 | 北京百度网讯科技有限公司 | Data processing method and device, electronic equipment and storage medium |
| CN113268350A (en) * | 2021-06-07 | 2021-08-17 | 上海数禾信息科技有限公司 | Resource allocation method and device based on cloud service construction and computer equipment |
| CN113268350B (en) * | 2021-06-07 | 2024-01-26 | 上海数禾信息科技有限公司 | Resource allocation method, device and computer equipment based on cloud service construction |
| CN113688989A (en) * | 2021-08-31 | 2021-11-23 | 中国平安人寿保险股份有限公司 | Deep learning network acceleration method, device, equipment and storage medium |
| CN113688989B (en) * | 2021-08-31 | 2024-04-19 | 中国平安人寿保险股份有限公司 | Deep learning network acceleration method, device, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112181659B (en) | 2023-07-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Nie et al. | Network traffic prediction based on deep belief network in wireless mesh backbone networks | |
| US10360517B2 (en) | Distributed hyperparameter tuning system for machine learning | |
| CN112181659B (en) | Cloud simulation memory resource prediction model construction method and memory resource prediction method | |
| US9864807B2 (en) | Identifying influencers for topics in social media | |
| CN111258767B (en) | Cloud computing resource intelligent distribution method and device for complex system simulation application | |
| CN112418482B (en) | Cloud computing energy consumption prediction method based on time series clustering | |
| CN106649479B (en) | Transformer state association rule mining method based on probability graph | |
| US9111227B2 (en) | Monitoring data analyzing apparatus, monitoring data analyzing method, and monitoring data analyzing program | |
| CN110858973A (en) | Method and device for predicting network traffic of cell | |
| CN113037877A (en) | Optimization method for time-space data and resource scheduling under cloud edge architecture | |
| CN111611488A (en) | Information recommendation method and device based on artificial intelligence and electronic equipment | |
| CN112580775A (en) | Job scheduling for distributed computing devices | |
| US11977993B2 (en) | Data source correlation techniques for machine learning and convolutional neural models | |
| CN110825522A (en) | Spark parameter self-adaptive optimization method and system | |
| CN113282409A (en) | Edge calculation task processing method and device and computer equipment | |
| CN113158435B (en) | Complex system simulation running time prediction method and device based on ensemble learning | |
| CN113094899B (en) | Random power flow calculation method and device, electronic equipment and storage medium | |
| Singham et al. | Density estimation of simulation output using exponential epi-splines | |
| CN112907124B (en) | Data link abnormity evaluation method and device, electronic equipment and storage medium | |
| CN115001978A (en) | Cloud tenant virtual network intelligent mapping method based on reinforcement learning model | |
| Iordache et al. | Predicting service level agreement violations in cloud using machine learning techniques | |
| CN117931420B (en) | Cloud workload prediction method, device, equipment and storage medium | |
| CN116611506B (en) | User analysis model training method, user label determining method and device | |
| Xinxiang | Collaborative filtering recommendation using matrix factorization: A mapreduce implementation | |
| CN117081942A (en) | Network traffic prediction method, device, equipment and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |